<h2>Must know</h2>
<ul>
<li><strong>Laravel 12 launches official AI SDK</strong>: Laravel 12 has officially released its native AI SDK, providing a unified interface for text generation, embeddings, and agentic workflows across providers like Anthropic and OpenAI. The SDK includes built-in support for structured outputs and automatic fallbacks, effectively standardizing how AI is integrated into the framework. According to Laravel News, this release signals a shift from experimental packages to AI being a core architectural pillar of the ecosystem. <a href="https://app.daily.dev/posts/fqJ4PUxyc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LiteLLM supply chain attack harvests developer secrets</strong>: A critical supply chain attack targeted the LiteLLM package on PyPI, using a compromised GitHub Actions workflow to inject credential-stealing malware. The malicious code automatically exfiltrates SSH keys, cloud tokens, and environment variables upon process startup. Wiz Research reports that the attack exploited a misconfiguration in the Trivy scanner’s CI/CD pipeline, highlighting a dangerous new vector where security tools themselves are weaponized to harvest secrets. <a href="https://app.daily.dev/feed-by-ids?id=xt7JWjaFB&amp;id=b5NsKzD82" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MCP gains momentum with Microsoft and WordPress adoption</strong>: The Model Context Protocol is rapidly becoming the industry standard for connecting AI agents to external data, with Microsoft and WordPress launching major integrations. Microsoft’s new Azure DevOps Remote MCP Server allows agents to securely query work items and pipelines via Entra ID authentication. Simultaneously, WordPress 7.0 has introduced a Connectors API to manage these AI-to-service relationships natively within the CMS. <a href="https://app.daily.dev/feed-by-ids?id=9yY7DGd7z&amp;id=6us9IvcEV" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Declarative data imports arrive for Laravel</strong>: Laravel Ingest provides a configuration-driven ETL layer for streaming massive CSV and Excel files using PHP Generators. <a href="https://app.daily.dev/posts/C9rpuhh7R" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Versioned prompt management for Laravel</strong>: Prompt Deck allows managing AI prompts as versioned markdown files on disk with support for variable interpolation. <a href="https://app.daily.dev/posts/HLhWYIKDD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare Workers AI provider for Prism PHP</strong>: Prism Workers AI adds Cloudflare Workers AI as a provider for the Laravel AI SDK, supporting multi-step tool calling. <a href="https://app.daily.dev/posts/vbnbCooLn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Static analysis for database schemas</strong>: DBStan performs static analysis on database schemas to detect missing indexes and normalization issues in Laravel apps. <a href="https://app.daily.dev/posts/K5sWdO7ie" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Native circuit breaker support for Laravel</strong>: The circuit-breaker package prevents cascading failures by wrapping downstream service calls with automated state management. <a href="https://app.daily.dev/posts/W00ae4UdJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>VS Code style syntax highlighting in PHP</strong>: Phiki brings TextMate-powered syntax highlighting to PHP, using VS Code themes to generate web-ready code blocks. <a href="https://app.daily.dev/posts/fLcf1VNKV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Standardized PDO provider interface proposed</strong>: A new PHP-FIG proposal introduces a PdoProviderInterface to abstract lazy connections and pooling across different DBALs. <a href="https://app.daily.dev/posts/Jf2ekk7K1" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The Cursor Composer 2 rebrand charade</strong>: Cursor recently launched its Composer 2 model with much fanfare, only for users to discover it’s actually a rebrand of Moonshot AI’s Kimi k2.5. While Cursor added some reinforcement learning on top, the ‘in-house model’ narrative looks a lot like license laundering through an inference partner. It’s a classic move: wrap an open-weight model in a fancy UI and hope nobody checks the API endpoints. <a href="https://app.daily.dev/feed-by-ids?id=0hBN6hU9R&amp;id=6IlX0UUsy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DLSS 5 is just neural makeup for bad optimization</strong>: Nvidia is trying to convince us that DLSS 5 is a ‘graphics breakthrough,’ but gamers are calling it out for what it is: an over-processed AI filter that smears textures and ruins artistic intent. The demo required two RTX 590s just to run, proving that we’re now using massive amounts of compute to fix the artifacts created by the previous layer of AI. We’ve officially reached the ‘hallucinating pixels to save performance’ stage of graphics engineering. <a href="https://app.daily.dev/feed-by-ids?id=BsIJ2t6dP&amp;id=P7dnxxeG8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Egg receipt automation and the death of ROI</strong>: The latest trend in ‘vibe coding’ involves developers spending $1,500 in LLM tokens to automate tasks that yield $1,900 in data, like the recent project processing 25 years of egg receipts. While technically impressive, it highlights the total lack of ROI discipline in the current AI boom. We’re burning through investor subsidies to build over-engineered pipelines for problems that a simple regex or a junior intern could have solved for a fraction of the cost. <a href="https://app.daily.dev/feed-by-ids?id=cJ6OaPAbu&amp;id=8CuxGxfHz" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
